An h-adaptive stochastic collocation method for stochastic EMC/EMI analysis

Abdulkadir C. Yücel, Hakan Bagci, Eric Michielssen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

6 Scopus citations


The analysis of electromagnetic compatibility and interference (EMC/EMI) phenomena is often fraught by randomness in a system's excitation (e.g., the amplitude, phase, and location of internal noise sources) or configuration (e.g., the routing of cables, the placement of electronic systems, component specifications, etc.). To bound the probability of system malfunction, fast and accurate techniques to quantify the uncertainty in system observables (e.g., voltages across mission-critical circuit elements) are called for. Recently proposed stochastic frameworks [1-2] combine deterministic electromagnetic (EM) simulators with stochastic collocation (SC) methods that approximate system observables using generalized polynomial chaos expansion (gPC) [3] (viz. orthogonal polynomials spanning the entire random domain) to estimate their statistical moments and probability density functions (pdfs). When constructing gPC expansions, the EM simulator is used solely to evaluate system observables at collocation points prescribed by the SC-gPC scheme. The frameworks in [1-2] therefore are non-intrusive and straightforward to implement. That said, they become inefficient and inaccurate for system observables that vary rapidly or are discontinuous in the random variables (as their representations may require very high-order polynomials). © 2010 IEEE.
Original languageEnglish (US)
Title of host publication2010 IEEE Antennas and Propagation Society International Symposium
PublisherInstitute of Electrical and Electronics Engineers (IEEE)
ISBN (Print)9781424449682
StatePublished - Jul 2010

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01


Dive into the research topics of 'An h-adaptive stochastic collocation method for stochastic EMC/EMI analysis'. Together they form a unique fingerprint.

Cite this